Autism spectrum disorders (ASDs) are a range of neurodevelopmental disorders affecting social communication and behavior. Brain network dysfunctions are supported as a neurobiological basis for ASDs. Thus, functional connectivity (FC) studies are pivotal for unraveling autism-related large-scale network dynamics. Although functional imaging studies during development can be instrumental for autism diagnosis, longitudinal studies of FC consolidation over development to adulthood are missing in mice. SHANK3 is a postsynaptic scaffolding protein of excitatory synapses. Its deletion or mutation is well-known to cause a rare genetic disorder named Phelan-McDermid syndrome, which is characterized by atypical sensory processing. In awake resting-state Shank3 mutant mice (Shank3b+/+ and Shank3b+/-), we longitudinally monitored cortical network alterations from post-natal day 45 (P45) to P90 using mesoscopic Ca2+ imaging of excitatory neurons. We found that hyper-connectivity of the barrel cortices plays a significant role in the emergence of aberrant FC patterns, starting at a juvenile age (P45). By leveraging whisker stimulation, we also revealed increased excitability of the stimulated barrelfield cortex associated with strong bilateral hyperconnectivity of the motor cortices in Shank3b+/- mice. Finally, we developed an artificial intelligence model based on a convolutional neural network for the automatic classification of autism deficits from the distributed cortical responses evoked by whisker stimulation. These findings highlight a key pattern of cortical dysfunction associated with autism and a potential early target for non-invasive translational treatments.